Heterogenising study samples across testing time improves reproducibility of behavioural data

Sylvia Kaiser,Fabian Karwinkel,S Helene Richter,Norbert Sachser,Vanessa Tabea Von Kortzfleisch,Carina Bodden

doi:10.1038/s41598-019-44705-2

Sylvia Kaiser, Fabian Karwinkel + Show 4 more

Open Access

https://doi.org/10.1038/s41598-019-44705-2

Copy DOI

Abstract

The ongoing debate on the reproducibility crisis in the life sciences highlights the need for a rethinking of current methodologies. Since the trend towards ever more standardised experiments is at risk of causing highly idiosyncratic results, an alternative approach has been suggested to improve the robustness of findings, particularly from animal experiments. This concept, referred to as “systematic heterogenisation”, postulates increased external validity and hence, improved reproducibility by introducing variation systematically into a single experiment. However, the implementation of this concept in practice requires the identification of suitable heterogenisation factors. Here we show that the time of day at which experiments are conducted has a significant impact on the reproducibility of behavioural differences between two mouse strains, C57BL/6J and DBA/2N. Specifically, we found remarkably varying strain effects on anxiety, exploration, and learning, depending on the testing time, i.e. morning, noon or afternoon. In a follow-up simulation approach, we demonstrate that the systematic inclusion of two different testing times significantly improved reproducibility between replicate experiments. Our results emphasise the potential of time as an effective and easy-to-handle heterogenisation factor for single-laboratory studies. Its systematic variation likely improves reproducibility of research findings and hence contributes to a fundamental issue of experimental design and conduct in laboratory animal science.

Highlights

Over the last years, a novel keyword has found its way into the scientific debate: the “reproducibility crisis”[1,2,3,4]
We used matched pairs of C57BL/6J and DBA/2N mice that lived in neighbouring cages, sharing the same microenvironment
The inability to reproduce research findings is increasingly perceived as a major problem[3,24]

Summary

Introduction

A novel keyword has found its way into the scientific debate: the “reproducibility crisis”[1,2,3,4]. The logic of this concept was supported by a series of proof-of-principle studies[22,23,25], there is still no effective heterogenisation strategy for a single-laboratory experiment Such a strategy would require systematic variation of experimental factors within a laboratory that interact with the treatment under investigation in a way that prevents the detection of idiosyncratic or lab-specific results. The aim of the present study was to evaluate the potential of testing time as a feasible and easy-to-handle heterogenisation factor for single-lab studies For this purpose, we systematically investigated the impact of three different testing times on the reproducibility of treatment effects. We hypothesised to observe improved reproducibility of behavioural strain differences in the heterogenised compared to the standardised design

Objectives

Methods

Results

Conclusion